1、APPG on Music report:Artificial Intelligence and the Music Industry Master or Servant?The All-Party Parliamentary Group on Music in association with UK Music.This is not an official publication of the House of Commons or the House of Lords.It has not been approved by either House or its committees.A

2、ll-Party Parliamentary Groups are informal groups of Members of both Houses with a common interest in particular issues.The views expressed in this report are those of the group.This report was researched and funded by UK Music.2ContentsForeword Kevin Brennan MP,All-Party Parliamentary Group on Musi

3、c Chair4What Do People Think About AIs Impact On Music?6Introduction 7Recommendations 8About the APPG on Music 11A Beginners Guide to AI and the Music Industry 12APPG on Music Inquiry Findings Consumer Protection 14Fair Market Access 16Voice and Image Likeness 18International Action 20AI in Action:D

4、AACI22Hopes and Concerns of the Music Industry on AI 25Conclusion 28Acknowledgements 30Glossary of Terms31Annex 32References334ForewordArtifical Intelligence(AI)is having a profound impact on our musical landscape.The UK must grasp the transformative potential of AI in shaping the future of music if

5、 it is to retain its role as a powerhouse in exporting music and in nurturing world-class talent.We must also confront the danger that unfettered developments in AI could pose to the UKs musicians and music businesses.We ignore the necessity to sow policies,which will harvest the benefits of AI,and

6、help stave off the threats it poses,at our peril.Our central insight must always be that AI can be a great servant but would be a terrible master.Artificial intelligence does not exist without human intelligence.AI doesnt create art in the human sense but ingests the patterns of human creativity to

7、generate music and other outputs based on algorithms and predictions.There is no ghost in the machine-no soul in the computer.AI is a desiccated calculating machine.But as a tech maestro,AI is capable of analysing,generating and even performing music in ways previously unimaginable.It can help music

8、ians to innovate and to inspire new human creativity.UK-based companies such as DAACI,who featured as part of this inquiry,are already thinking of imaginative new ways to bring together music and tech to ensure that human creativity is always recognised and rewarded.By leveraging the collective stre

9、ngth of policymakers,industry leaders,and innovators we can ensure that AI serves as a catalyst for creativity and progress in the music ecosystem,rather than an inhibitor of growth and a destroyer of creators livelihoods.I am delighted to present this cross-party report,which outlines a strategy fo

10、r embracing the opportunities in music presented by AI,while safeguarding the integrity of our music economy.I believe that my parliamentary colleagues have used their human intelligence to develop findings and recommendations that provide a clear blueprint for the future use of artificial intellige

11、nce in music making.Our report is underpinned by an overarching legislative ask for the Government to create a UK AI Bill.Such a Bill should be pro-copyright and pro-creative industries.Kevin Brennan MPAll-Party Parliamentary Group on Music ChairThe proposed law would be a vehicle to ensure creators

12、 and consumers are protected in this new environment,through updated robust copyright,clear labelling,accurate record keeping and new personality rights.It would also ensure the UK is not left behind internationally as other nations move to legislate.Intellectual property is a human concept.There is

13、 no such thing as artificial intellectual property.Copyright law exists to incentivise human creativity by rewarding humans for the use of their intellectual property.AI companies cannot simply pocket that property and repackage it for their own gain without permission or a licence from its owners.T

14、here have been many technological developments since music was first printed on a page,punched onto a pianola roll,recorded on wax cylinder,disc and tape,or digitised on a computer.AI presents perhaps the greatest challenge of all,but each technological advance also generates an opportunity for crea

15、tors.We need to muster the will and the vision to seize this one.7What Do People Think About AIs Impact On Music?83%of UK adults agree that if AI has been used to generate a song it should be clearly labelled.2 OUT OF 3adults in the UK are concerned about the risk of AI generation eventually replaci

16、ng human creativity.77%of UK adults agree that AI-generated music that does not acknowledge the original musics creators amounts to theft.4 OUT OF 5UK adults agree that the law should prevent an artists music from being used to train an AI application without their knowledge or permission.Recent adv

17、ancements in AI have been as rapid as they are transformative.We are already witnessing transformation and change in real time,with platforms such as OpenAI,which runs ChatGPT,and chatbots making AI more accessible and popular than ever before.Just as we have seen with the growth of the internet,the

18、se technological developments bring huge opportunities,but also massive risks and challenges without the right guardrails in place for its development.The music industry is,and will always be,a highly innovative industry,which embraces and uses new technologies to create and consume music.The indust

19、ry is already taking advantage of the power of AI and has been for some time.Creators use it as a tool assisting their artistic endeavours,in the song writing process,mastering their recorded performance,improving their sound,and gaining new insights into fan engagement and fanbases around the world

20、.Equally at a business level,music publishers and record companies use AI to assist their administration and enforcement operations.The discussion on AI in the music industry heightened in July 2022,when the UK government announced that it intended to introduce a text and data mining exception to co

21、pyright to support the development of AI.The proposals suggested an amendment to copyright law that would enable developers of AI to exploit copyright protected works without the permission of creators and rightsholders.The new copyright exception received a huge backlash from the UK music industry,

22、which likened the plans to music laundering and warned of a catastrophic impact on the sector.After months of campaigning by the music and creative industries,in January 2023 in a debate in the House of Commons the then intellectual property minister George Freeman,supported by this All-Party Parlia

23、mentary Group(APPG),said he and the Department for Culture,Media and Sport(DCMS)minister Julia Lopez felt that the proposals were not correct and that they would not proceed with them after all1.Whilst seen as a positive step,the industry is still concerned that the rights of creators are being expl

24、oited by tech companies and there is a need to push for further legislation to protect music creators and rightsholders creative property.Consultation continues with government,but progress is slow and alternative proposals for a voluntary cross-industry code of conduct stalled in 2023.In recent mon

25、ths several creators have expressed their concerns on the growth of AI.Jamie Cullum,Jess Glynne,Mumford and Sons,Robert Smith,Sam Smith,The Last Dinner Party,Yard Act and Zayn Malik are among the UK acts who signed an open letter calling for protections against the predatory use of AI.The letter,iss

26、ued by the US-based Artist Rights Alliance advocacy group,made the demand that technology companies pledge not to develop AI tools that undermine or replace human songwriters and artists2.This is not just a music industry concern;the general public are worried too.As polling commissioned by UK Music

27、 published in this report shows3,four out of five(80%)of UK adults agree that the law should prevent an artists music from being used to train an AI application without their knowledge or permission.Another four out of five(83%)of UK adults agree that if AI has been used to generate a song,it should

28、 be clearly labelled.Over two-thirds(69%)of UK adults are concerned about the risk of AI generation eventually replacing human creativity.The figures tell the same story as global polling commissioned by the International Federation of the Phonographic Industry(IFPI)which found 79%of people feel hum

29、an creativity remains essential to the creation of music,76%feel that an artists music or vocals should not be used or ingested by AI without permission and 74%agree that AI should not be used to clone or impersonate artists without authorisation4.This All-Party Parliamentary Group(APPG)on Music rep

30、ort presents nine crucial recommendations for the UK government,across the four key themes,which emerged from the inquiry sessions:consumer protection,fair market access,protection of voice and image likeness,and a need for international action.Underpinning these recommendations is a call to introdu

31、ce an ambitious Parliamentary Bill on AI which supports the creative industries.Such legislation would seek to uphold the UKs strong copyright standards through the 21st century.It would act as a vehicle to implement many of the recommendations included in this report and called for by the cultural

32、sector.Such a Bill would ensure the UK does not get left behind in the race to prosper from developments in AI.While other countries are providing a patchwork approach to regulating generative AI,the UK could be world-leading on producing single,coherent and comprehensive AI Bill.Recommendation 1:Th

33、e Government should create a pro-creative industries AI Bill.As well as protecting copyright,the Bill should introduce new rights and obligations around labelling and record keeping as well as enhancing personality rights.It should act as a vehicle to implement many of the recommendations of this re

34、port and show the UK can lead the way in harnessing new technology.IntroductionPolling was commissioned by UK Music and conducted by Whitestone Insight.Whitestone Insight surveyed 2,110 adults in the UK online between 20th and 21st March 2024.Data were weighted to be representative of all UK adults.

35、Whitestone Insight is a member of the British Polling Council and abides by its rules.Full tables can be found on Whitestone Insights website.References can be found on page 3389ReportRecommendationsBelow is the full list of recommendations from APPG on Music.Throughout the report an explanation of

36、the evidence is provided,followed by the relevant recommendation.Recommendation 1:The Government should create a pro-creative industries AI Bill.As well as protecting copyright,the Bill should introduce new rights and obligations around labelling and record keeping as well as enhancing personality r

37、ights.It should act as a vehicle to implement many of the recommendations of this report and show the UK can lead the way in harnessing new technology.Recommendation 2:Transparent labelling enables informed decision-making for consumers.The Government should ratify labelling requirements for AI-gene

38、rated content under the Consumer Rights Act 2015.Recommendation 3:The Government should introduce a standalone obligation for AI developers and those using Large Language Models(LLMs)to comply with record keeping requirements for all data sets used for ingestion,not solely limited to personal data.R

39、ecommendation 4:The Government should promote compliance with UK copyright law,requiring stakeholders to obtain express permission before using copyrighted material,and educating music creators and rightsholders on their rights.Recommendation 5:The Government should address the copyright status of A

40、I-generated works and provide clarity,making it clear that without human creativity copyright is not afforded as a right.Recommendation 6:The Government should introduce a specific personality right to protect creators and artists from misappropriation and false endorsement.Such a right should prote

41、ct their voice,image,name,and likeness(VINL).Recommendation 7:As a condition of market access,the Government should require Large Language Models(LLMs)to comply with UK copyright provisions,even if the services or goods they have developed are created in compliance with the laws outside the UK.This

42、could also be achieved by clarifying that when an individual or organisation create goods or services internationally which breach copyright law in the UK,that UK law would apply.Recommendation 8:The UK Government should take the lead role in the creation of an international taskforce on AI,taking f

43、orward this dialogue across borders,ensuring best practice is shared and putting copyright and the creative industries at the heart of any future global AI summit.10About the All-Party Parliamentary Group on Music The All-Party Parliamentary Group(APPG)on Music exists to provide a forum for parliame

44、ntarians to discuss issues of concern to them and their constituents.APPGs bring together MPs and Peers from across the party-political divide and act as a point for dialogue amongst decision-makers on issues of concern.Through a series of events and discussions during the political year,the APPG on

45、 Music seeks to inform parliamentarians about developments within the music industry as well as engage MPs and Lords on measures that impact the sector.UK Music provides the secretariat for the APPG on Music.The APPG on Music is one of the largest in Westminster with more than 100 MPs and Peers as m

46、embers.About the InquiryThe inquiry into AI was launched at an APPG on Music meeting in January 2024.At the meeting,the APPG discussed the organisations or individuals from whom they would seek evidence.The APPG were particularly interested in considering the cross-border impact of AI,with a desire

47、to hear from representatives from the EU and the US.The group wanted to hear a diverse range of voices to understand different angles to the debate.They also wanted to meet creators who are using AI and see it in action,as well as learn more about the concerns of the industry.Some members of the APP

48、G admitted they knew very little about AI but were keen to learn more.Other members were already highly engaged with the topic.The APPG on Musics House of Lords spokesperson,Lord Clement-Jones,released his own,comprehensive book on AI during the inquiry.Titled Living With the Algorithm:Servant or Ma

49、ster?the book looks at the potential future governance of AI in the UK5.This range of expertise provided a helpful backdrop to the inquiry,including guiding the types of panellists we invited,and the questions asked.Formal evidence sessionsThe APPG received evidence from individuals and organisation

50、s with experience across the UK,EU,and the US.The first two sessions were run in line with a UK select committee inquiry,with the APPG asking questions of industry experts.The third session featured an in-person demonstration on AI.The sessions were:On February 27,2024,the first panel took place.The

51、 APPG on Music heard from UK Musics legal advisor Florian Koempel;Professor of Intellectual Property and Technology Law at Queen Mary University of London Noam Shemtov and The European Authors Societies(GESAC)General Counsel,Burak zgen.The session focused on progress and lessons so far from the EU a

52、nd its AI Act.On March 22 2024,Shira Perlmutter,Register of Copyrights and Director of the US Copyright Office and her team presented a view from the US to the APPG on Music in the second session for the APPG on Music.On March 25,2024,AI tech firm DAACI offered a practical presentation,featuring dem

53、onstrations of DAACIs patented AI tools.DAACI CEO Rachel Lyske and the DAACI team presented an overview of the integration of AI in music creation in the final session of the APPG on Music inquiry.Throughout the report,we highlight verbal evidence from sessions that relate to the topic.Written evide

54、nce and industry engagement As part of the inquiry,the APPG welcomed written evidence from the public,which was sought through UK Music and its members on social media(full questions in annex one).These submissions helped guide the report content,and quotes from these are provided throughout this re

55、port.Organisations from across the music industry also contributed to the report by answering the question;“What is your main hope and your main concern about AI?”.Answers can be found on page 25.The inquiry panels,sessions,written evidence and industry engagement helped guide the four key themes th

56、at came out of the inquiry:Consumer protection,fair market access,voice and image likeness and international action.References can be found on page 331213A Beginners Guide to AI and the Music Industry Operation of AI machines The first element the APPG on Music wanted to understand was how does AI a

57、nd music work.The Parliamentarians heard evidence from six different experts to properly understand the practicalities.To note:when discussing AI,we are most often talking about generative AI rather than assistive AI.AI applications are software programmes that provide an answer to a question based

58、on probabilities.In its most simplified terms,AI systems will have learned from training datasets that the most likely letter following an A is an N based on the frequency of letter pairs in the English language.The most probable next letter would be a D(as in and,hand,land and so on).The more lette

59、rs and words ingested,the more complex the algorithmic process becomes,and the better the probabilities.This is called a large language model(LLM).Ultimately this is a question of computational resources.AI applications need to retain the datasets to understand context and to operate efficiently,as

60、words have different meanings in different contexts.The real artificial world is more complicated,depending on the actual model used e.g.diffusion or transformer.Given that computers do not understand lyrics or music,the AI application first needs to translate the music or the lyrics or music into t

61、heir binary language before identifying patterns,relations,and context in a process called deep learning.Deep learning can be compared to a human who is stung by a bee learning that black and yellow striped insects are hurtful(ingested data)and should therefore on balance be avoided(inference).Funni

62、ly enough,some insects have evolved to pretend to be hurtful by mimicking black and yellow stripes(in essence,the animal version of a deepfake).The more data LLMs ingest,the more likely the probability of producing accurate,useful output.ChatGPT 4.5 in its most recent application has been trained on

63、 45 gigabytes of data;its predecessor GPT 3 only had 17 gigabytes.Ultimately,the key factor for the success of an AI application is the quality and quantity of the data.In our case,the songs created by humans ingested by a machine.A simplified definition of AI:Algorithms use existing datasetsInputTo

64、 make predictions based on probabilitiesOutputMimicking human creativityTo produce these outputs,AI services and platforms scrape the internet to collect and provide data to AI applications.This involves many rights that require express permission,including copyright.Such requirement is not supersed

65、ed by any of the potentially available exceptions including text and data mining,temporary copying or fair use depending on the jurisdiction.In any case,an exception would only apply to copyright,but not to other rights such as data protection and unfair competition rules.Additionally,data scraping

66、is often expressly prohibited in the terms and conditions of the scraped websites;this constitutes a legally binding express prohibition which needs to be respected.This is AI in its most simplified definition,but hopefully helps set the scene of how some issues arise from AI for the music industry.

67、1415APPG on Music Inquiry Findings:Consumer ProtectionConsumer choiceA clear theme from the inquiry is that AI and music is as much a consumer issue as it is an industry issue.We heard that while some consumers might prefer AI-generated works,others might prefer human-created works,depending on thei

68、r individual philosophy regarding technology,costs,environmental impact and other factors.“It should be good practice to keep records,but more importantly,the labelling.The consumer should have a clear choice of what product they want.If I want to buy fair trade clothing.I need to be assured that it

69、s fair trade”.Florian Koempel on the importance of labellingThe APPG heard that in the same way some consumers acquire fair trade clothing because of their views on sustainability,and others might not care or might not be able to afford fair trade clothing,consumers should be able to make an informe

70、d decision based on transparent information about the product they are acquiring.UK Musics polling revealed over four out of five UK adults agree that if AI has been used to generate a song,it should be clearly labelled.Over half of UK adults are concerned about listening to AI-generated music witho

71、ut realising it.Labelling At the core of enabling consumers to make this choice,labelling is key and should be prescribed,including as part of the UK consumer protection framework.Labelling AI-generated products allows consumers to make informed decisions.It also means human creators will be properl

72、y recognised for their intellectual contributions.In the context of music,labelling can be as simple as a requirement within the metadata to identify that music is AI-generated.“As with any product in the marketplace,origin of goods is key to protect consumers.Music and pop culture has the power to

73、influence it is important for a consumer to know whether material has been generated by a computer,or if it is authentic human creation”.Creative Artists Agency6 on the power of pop cultureA legislative solution would be to amend the Consumer Rights Act 2015 to introduce labelling requirements for A

74、I-generated content.Using this as the parent act underscores the APPG on Musics clear evidence that there is a consumer protection dimension to labelling requirements for AI-generated music.Recommendation 2:Transparent labelling enables informed decision-making for consumers.The Government should ra

75、tify labelling requirements for AI-generated content under the Consumer Rights Act 2015.Record keeping Through the inquiry,it became obvious that to label products properly tech providers must keep an auditable record of the music used at the input stage.This includes associated content and any meta

76、data ingested before the algorithm generates new music or other content.Crucially,the input stage is the only point in the process when these data points can be documented.Without accurate records creators and rightsholders are at real risk of not being credited or remunerated for their work.“You do

77、nt know whether or how those companies have used your work,because the output generated by AI technologies,as they get more sophisticated,do not include identifiable pieces of original works that they are trained on.you dont know which specific works are in the datasets or to what extent they are be

78、ing used to train the model or to generate certain output”.Burak zgen on the difficulties of knowing what work has been usedTechnology in this area is moving rapidly.As our inquiry panellists highlighted,it can be difficult to know whether an individuals work has been used or not,but once content is

79、 ingested it is near impossible to remove it from the process.While improvements in the future may help track the use of music and other data,it would be a big mistake to rely on this to solve the issues of today.Its imperative to introduce a requirement as soon as possible for tech providers to com

80、ply with record-keeping requirements.Requirements for record keeping are beginning to be included in regulatory AI proposals across the world,including the EU and China.Without improvements in the UK,we are at serious risk of being left behind.The Information Commissioners Office(ICO),the UKs indepe

81、ndent regulator for data protection and information rights,is currently establishing its regulatory role on generative AI and appears ideally placed to supervise a specific standalone provision.Recommendation 3:The Government should introduce a standalone obligation for AI developers and those using

82、 Large Language Models(LLMs)to comply with record-keeping requirements for all data sets used for ingestion,not solely limited to personal data.55%of UK adults are concerned about listening to AI generated music without realising it.4 OUT OF 5UK adults agree that if AI has been used to generate a so

83、ng it should be clearly labelled.Full tables can be found on Whitestone Insights website.Full tables can be found on Whitestone Insights website.16Creators choice In the inquiry,the APPG on Music heard of the inevitability that AI-generated works will compete with human-created works in the future.T

84、hey will coexist and cater for different needs.Therefore,it is crucial they compete on a level playing field.“When it comes to creators choice,thats a critical point,because creators want to be able to give consent and receive remuneration for the use of their works.Thats what we need to ensure”.Bur

85、ak zgen on the importance of creators choiceLike human residents in the UK,AI application providers must comply with the law of the land.An AI application should not generate a product consisting of several other works ingested without permission and subsequently be allowed to compete with the origi

86、nal works.This means the creators and their rightsholders should be able to decide if and how they want to use their creative talent to be used.AI developers need to respect creators choice as a baseline for any discussions.UK Musics polling found 82%of UK adults agree that it should be up to artist

87、s or their chosen rightsholder to decide whether their music is used as the basis for creating new AS-assisted compositions or tracks which others can then profit from.77%of UK adults agree that AI-generated music that does not acknowledge the original musics creators amounts to theft.Anyone using t

88、he music protected by copyright needs to obtain express permission in the absence of an applicable exception.Notwithstanding pending legal cases,this is an opportune moment for the Government to encourage compliance with the existing UK copyright framework.This should be supplemented by additional e

89、ducation for music creators and artists on their rights.Recommendation 4:The Government should promote compliance with UK copyright law,requiring stakeholders to obtain express permission before using copyrighted material,and educating music creators and rightsholders on their rights.Exception and o

90、pt-outs As acknowledged in the APPG inquiry sessions and by the former IP Minister George Freeman in 2023,a new copyright exception removing the rights of UK creators and artists to consent would not have been appropriate.An exception with an opt-out mechanism,as chosen by the EU,does not effectivel

91、y provide relief for creators and artists.The practical application of the opt-out mechanism throughout the EU demonstrates significant problems in practice.Uncertainties surrounding the person opting out,the person receiving the opt-out,the instrument opting out and transparency regarding opt-out c

92、ompliance,render the process useless,in particular for individual creators and artists.“It is very far from clear what is covered by fair use in the US.The system in the UK is slightly different,and it is much better to think about the issue in advance and decide what is the right path for the UK,ra

93、ther than rely on fact-specific judicial decisions as in the case of the fair use”.Noam Shemtov on deciding what is best for the UK“If copyright owners are given the ability to either opt in or opt out of inclusion in the training materials,how can that work,especially for individual creators who ar

94、ent as sophisticated as corporations?If consent or compensation is required,how can that feasibly be accomplished,given the volume and variety of the works involved?”.Shira Perlmutter on some of the questions associated with the opt-outFurthermore,AI platforms will not be able to consider all opt-ou

95、ts in an economically feasible way.Due to the nature of their machine learning processes,it is impossible to identify the specific written content learned from an opt-out work.To properly consider the opt-out,the AI service providers would need to repeat the whole ingestion process from scratch.This

96、 is not practical for AI platforms.Equally,creators and artists are not able to transparently check whether an AI platform has respected the opt-out.APPG on Music Inquiry Findings:Fair Market Access Copyright for human creativity?AI-generated works,or AI-generated elements of works,are mostly not pr

97、otected by copyright;they lack key conditions of copyright protections across different territories.This is evident across different legal cases:In the UK,to benefit from copyright protection,the stamp of individuality reflective of the creation of the author(s)is required.Notably,in a recent High C

98、ourt decision relating to patent law,the court stated that the inventor must be a“natural person”7,8.In the EU,copyright protects the authors own intellectual creation expressing their personality.That condition still applies in the UK despite leaving the EU,according to a 2023 legal decision9,10.In

99、 the US,the situation is even more direct.The US District Court for the District of Columbia stated that there is no copyright for a work generated entirely by an artificial system without human involvement.Furthermore,the US Copyright Office compendium expressly states that it does not register wor

100、ks created by machines11.“If its AI-produced,you cant copyright,because copyright has to be a human part of human creativity,and it exists for that purpose to reward human creativity and encourage it as a concept”.Kevin Brennan MP on the purpose of copyrightIn addition to these legal cases,in the UK

101、 a specific law exists for computer-generated works,Section 9(3)of the Copyright,Designs and Patents Act(CDPA)1998.While this is a seemingly relevant provision it is barely understood,irrelevant in practice and only referenced in passing in one legal case two decades ago.It is therefore debatable wh

102、ether this Section has any practical value in this scenario.Recommendation 5:The Government should address the copyright status of AI-generated works and provide clarity,making it clear that without human creativity copyright is not afforded as a right.Full tables can be found on Whitestone Insights

103、 website.References can be found on page 331819APPG on Music Inquiry Findings:Voice and Image Likeness Personality rights The APPG on Music inquiry sessions highlighted how key it is to protect the voice,image,name,and likeness or VINL-of human creators from AI-generated output.These AI-generated cr

104、eations that mimic real people are sometimes known as deepfakes,and are becoming more prevalent across music,politics and wider society;often to spread misinformation or hurt individuals.“Performing artists,especially,have been raising the alarm about AI-generated works that imitate their likeness,s

105、tyle,and voice.Weve all seen the publicity about deep fakes including the AI-generated duet that appeared to be by Drake and the Weeknd.AI-generative versions of Tom Hanks and Taylor Swift have also been used to sell products,or for less savoury purposes”.Shira Perlmutter on the recent,high-profile

106、examples of deepfakes“Probably the strongest example there is of publicity right in the United States.is in the state of Tennessee and the reason is Graceland.They have a financial interest in protecting anything and everything that has to do with Elvis Presley”.Noam Shemtov on the strongest example

107、 of personality rights in the USThe US has been particularly quick to monitor and reduce potential damage in this area.Also in the US,in March 2024,Tennessee became the first state to enact legislation directly intended to protect musicians from having their vocal likeness generated by AI for commer

108、cial purposes.The Ensuring Likeness,Voice,and Image Security Act,or ELVIS Act,will be in effect from 1 July 2024 and makes it illegal to replicate an artists voice without their consent.In our inquiry sessions,the APPG on Music heard about how Tennessee has been quick to act because of their financi

109、al and tourism interest in protecting the brand of Elvis Presley.“Currently,in the event of an AI-generated deepfake or voice clone,to bring a claim against the infringer in the UK,rightsholders are likely best trying to make a claim based on a range of rights,including“passing off”.The common law a

110、ction for passing off has been used to prevent false celebrity endorsements,but it remains to be seen whether it is flexible enough to take effective action against deepfakes”.Jonathan Coote12 on the current UK law on personality rightsIn the UK,a variety of instruments exist to potentially protect

111、the rights of individuals from misappropriation or false endorsement,including passing off rights.But as written evidence to the inquiry highlighted,it is yet to be seen if this would be effective against deepfakes.Unambiguous legislation that protects creators and artists from misappropriation and

112、false endorsement would provide clarity and certainty for all involved,including tech providers.Polling from UK Music found that 83%of UK adults agree that a music artists creative“personality”should be protected in law against being copied using AI.62%of UK adults are concerned about the rise of so

113、-called deep fakes of their favourite music artists.The protection of the human personality can be achieved through the introduction of a specific publicity or image right protecting creators and artists from misappropriation and false endorsement.This should protect their likeness,including their v

114、oice if distinctive and widely known.The resulting damages can be economic or otherwise,such as damage to the artists reputation,goodwill,or otherwise causing distress.Recommendation 6:The Government should introduce a specific personality right to protect creators and artists from misappropriation

115、and false endorsement.Such a right should protect their voice,image,name,and likeness(VINL).References can be found on page 33 Full tables can be found on Whitestone Insights website.20APPG on Music Inquiry Findings:International Action International standardsThe first two APPG on Music inquiry sess

116、ions both came back to the same question:how can we deal with these issues on a global scale?While discussing issues through the lens of UK legislation is clearly an important first step,it will not deal with all the issues we are currently presented with across the borders where creators,rightshold

117、ers,tech providers and policymakers intersect.“The whole discussion needs to be more international than national”.Florian Koempel on taking a more international approachFor example,it is expected AI service providers will outsource the machine learning process to territories which provide copyright

118、exceptions to avoid having to ask for permission.We require adequate legal protection for the importation of AI applications which would constitute a breach of UK law if carried out here.“We need to make sure it would not be beneficial to rely on international AI training havens when you offer AI to

119、ols in the UK”.Noam Shemtov on ensuring the UK is protected from training models internationallyIn addition to specific legal provisions in copyright law,general market access standards should apply.Such standards exist in other sectors such as pharmaceuticals,food,and drinks,and should be easily tr

120、ansferable to the creative sector.The Government must introduce specific standards for LLMs,which operate and generate revenue in the UK as a condition of market access.As part of such standards,LLMs would need to comply with UK copyright provisions,notwithstanding whether their services or goods wo

121、uld have been created in compliance with the local rules of a third-party jurisdiction.“There is certainly a need to have transparency and compliance with copyright rules,and to ensure them with this type of extra-territoriality.I think it would help creators across the world to have convergence,at

122、least on these”.Burak zgen on the importance of thinking extra-territoriallyRecommendation 7:As a condition of market access,the Government should require Large Language Models(LLMs)to comply with UK copyright provisions,even if the services or goods they have developed are created in compliance wit

123、h the laws outside the UK.This could also be achieved by clarifying that when an individual or organisation create goods or services internationally which breach copyright law in the UK,that UK law would apply.International cooperationThe APPG on Music inquiry discussed the importance of talking and

124、 working with one another internationally.The APPG believes the UK government should take the lead on the creation of an international taskforce on AI,taking forward the dialogue across different borders and ensuring best practice is followed.At the heart of this taskforce must be a pro-copyright an

125、d pro-creative industries ethos.Following the AI Summit hosted in the UK in 2023,where the creative industries were notably not invited,the UK should also look to put the creative industries at the heart of any future summits or international events.“Were aware that these are still early days,and it

126、 remains to be seen how these laws are going to be interpreted and applied in this new context.Clearly the approaches are not going to be identical,and they dont need to be identical,but a high level of compatibility would be desirable for all of our stakeholders.The more we all talk to each other,t

127、he better”.Shira Perlmutter on the importance of working across bordersRecommendation 8:The UK Government should take the lead role in the creation of an international taskforce on AI,taking forward this dialogue across borders,ensuring best practice is shared and putting copyright and the creative

128、industries at the heart of any future global AI summit.62%of UK adults are concerned about the rise of so-called deep fakes of their favourite music artists.Full tables can be found on Whitestone Insights website.2223In the final session from the APPG on Music inquiry,members of the group received a

129、 practical presentation from AI-tech firm DAACI.This allowed the group to see AI in action.About DAACI DAACI develops next-generation smart and AI creative music tools.Their series of patented technologies empower music makers to meet the rapidly growing demand for innovative ways to create music.Th

130、eir technologies encompass tools that are designed to supercharge the creative process,dynamically composing new music in real-time,and smart editing systems that seamlessly adapt existing tracks.Built by a world-class team of musicians and composers,DAACIs technology is based on over 30 years of re

131、search.Incorporating a growing portfolio of 75 granted patents and supported by partnerships with the UKRI Centre for Doctoral Training in Artificial Intelligence and Music at Queen Mary University of London and Abbey Road Reds The Incubator.About the demoThe session for the APPG on Music involved a

132、 demonstration of DAACIs patented AI tools and products made for musicians.It allowed APPG members to gain insights into the kinds of assistive technology musicians can use to enhance their creative process via a hands-on presentation and discussions with the DAACI team.CEO Rachel Lyske and the DAAC

133、I team presented an overview of the integration of AI in music creation.They also addressed industry concerns in the generative AI music space and explained how DAACIs technology protects and rewards creators.Asking DAACI:what should Government do?“Recognise the opportunity in front of you.Music isn

134、t just a file to download.Its a whole industry.Its teachers,musicians,studio engineers,artists,producers and performers,homegrown creators and major music technology companies,festivals,venues,fans.Music is all of us.Protect this economy and allow it to thrive.The UK is a world leader and exporter o

135、f quality music,recording innovation and world-class artists.There is a window of opportunity for the UK to also be the world leader and exporter of generative AI.These two things do not have to be separate.The UK government must and can give the homegrown UK music and UK music technology industry a

136、 chance to get it right.We can provide a high-quality,fair,human-led AI system that protects human artistry and acknowledges every part of the value chain if we demand that approach together.Do not be railroaded by big technology companies pushing their own agenda saying that attribution and artist

137、recognition cant be done.Do not accept it doesnt work like that.Do not accept an ask for forgiveness later approach.It can be done legally,ethically and properly and DAACI has shown it can”.Rachel Lyske,DAACI CEOAI in action:DAACIAPPG on Music Chair Kevin Brennan MP with the DAACI team2425AIMIndepen

138、dent music has a long history of innovation and adapting to new technologies,creating new deal structures to ensure mutual benefit and fair returns to artists.Sadly,many generative AI models seem to have decided to scrape first and apologise later.An internet search will reveal myriad AI engines tha

139、t have“trained”on anything they can find online,whether or not within the terms of use of legitimate sites and platforms.This will continue to happen until governments ensure that copyright works,data and artistic identities are protected,with international collaboration,a clear and cost-effectively

140、 enforceable framework and serious penalties to deter bad actors from ruining AIs creative potential.Our community already engages with some great AI tools and we look forward to working with more developers who are as passionate about music as we are,who prioritise human artistry and respect copyri

141、ght,build in transparency and build out bias,and who develop to promote public safety and shared success.BPIAI offers tremendous opportunities to the music industry when used to support the creative process.It also presents real and pressing risks to creators and rightsholders if our world-leading i

142、ntellectual property rights are undermined.To support human artistry and ensure that our successful industry can continue to grow and add value to the economy,we need a clear commitment from AI developers and policymakers that training using music protected by copyright and the rights of individuals

143、 requires permission and payment.We need acknowledgement that its unacceptable to take someones music or identity without their consent.AI companies must keep detailed records of inputs so rightsholders can verify them and,if licensed,ensure creators are remunerated.AI-generated content should be cl

144、early labelled.These measures will ensure that artists experience the benefits of AI and we maintain the flow of investment into human-created music:this is central to supporting future talent and the success of our industry.FACThe music industry,and in particular,artists have always been at the for

145、efront of emerging technologies.Indeed,today,artists are driving forward the use of AI technologies in exciting and innovative new ways.However,as the AI revolution enters the mainstream,it is essential that measures are taken to ensure that these powerful technologies are not used in a manner that

146、undermines human creativity.This means making sure that artists must retain control,provide consent,receive fair remuneration and be credited for use of their works,and rightsholders must seek this permission for AI-related uses.Furthermore,the UK government must catch up and take steps to ensure th

147、e law is extended to protect artists image,likeness and identity as well as their work.Ivors AcademyMusic makers have always been at the forefront of using new technologies in their craft.The Beatles explored the potential of multitrack recording.MIDI and digital sampling were crucial to the develop

148、ment of hip-hop in the 80s.AI is no different.As the representative organisation for songwriters and composers,we want to ensure that AI can be used by creators to explore new genres and creativity but is not being developed at the expense of creators and their rights,upon which their future ability

149、 to create relies.To ensure that AI works in service of human-originated creativity and not against it,it is imperative to reinforce that any use of copyright works to train AI and generate derivative work requires consent and the remuneration of songwriters and composers.To support this goal,creato

150、rs rights must be strengthened,and a new personality right created to protect the personality and likeness of songwriters and artists.Hopes and Concerns of the Music Industry on AI 26PRS for MusicAI has the potential to transform the music industry,bringing with it great opportunities and great chal

151、lenges.Many music creators have integrated AI into their working lives,finding new ways to enhance their creativity.But enthusiasm is often surpassed by concerns that AI-generated works will directly compete with human-created compositions,negatively impacting creators livelihoods.The assumption tha

152、t our collective musical history is merely another dataset for consumption by AI must be unequivocally opposed.Now more than ever,copyrights core tenets of authorisation,and moral and economic rights,remain vital:after all,that is what it means to own ones work.Transparency is also essential to thos

153、e principles being upheld.The onus must be on tech firms to produce accessible information so that creators can monitor the use of their works and,where necessary,enforce their rights.The pathway to a robust framework for the safe development of AI is not an aspiration,it is a necessity.PPLThe music

154、 industry is only too ready to embrace the power that can be delivered by artificial intelligence and were already seeing some benefits from AI such as improving workflow efficiencies.However,when it comes to generative AI,these tools need to be used with legitimacy,and with respect for the labours

155、of those who have fuelled them.It is critical that the rights of those who invest their time,talent and money into making music are protected.Regulators need to spend less time worrying about AI of the future and focus on the actual harms taking place today.Technology companies need to seek appropri

156、ate permissions and pay appropriate value to the creators whose very essence fuels their large language models.Legislators must create an environment where legitimate AI companies can show their compliance with a fair business model.A Fairtrade for AI,if you will.Musicians UnionThe MUs main hope is

157、that the Government introduce a new unwaivable individual creator consent requirement for AI training,separate to existing performers rights.This would give creators control of their creations in relation to AI training and prevent rights historically granted to rightsholders for the physical and di

158、gital reproduction of music to sell to the public,long before AI training was even known or understood,being used to grant licences to tech firms without consultation,to the potential detriment of the creator.Our main concern is the use of creators performances and compositions to train AI(whether t

159、hat use be through unregulated and unlicensed scraping of content by tech firms,or by agreed licensing of works by rightsholders without contemporary,specific consent being first gained)and the effect the resulting output of AI systems may have on future work opportunities for music creators.MPGAs a

160、 creator personally and as the representative for Music Producers,Mixers and Recording Engineers through the Music Producers Guild(MPG)we remain hopeful about the innovative possibilities AI brings to the table,offering new tools and techniques to enhance our creativity and efficiency.However,there

161、are crucial concerns that demand attention.Protection of music-makers publicity,personality,and personal data rights is paramount,requiring robust legislation and international collaboration.Technology companies must uphold copyright laws and secure explicit consent from individual music-makers befo

162、re using their work to train AI models.Fair compensation and transparency are essential,and creators should share in the ongoing financial rewards of AI-generated music that their work helped generate.Finally,clear labelling of creator-approved AI-generated works and transparent record-keeping of da

163、tasets is crucial for accountability and integrity in the industry.In embracing AI,we must ensure it enhances,rather than undermines,the rights and contributions of music creators.MPAGranting copyright for AI-generated music would provide a material risk of displacing human-created music,and lead to

164、 market distortion based on likely oversaturation by machine-generated music.The dilution of revenues payable for the use of copyright-protected works by the inclusion in that pool of billions of works generated at the press of a button would have a devastating impact on the businesses supporting th

165、e ecosystem that sustains creators.As has been recognised in many jurisdictions as diverse as the European Union and China,transparent record keeping is vital for a secure and confident operation of AI applications.Knowledge of the ingested musical/literary works is key for potential remuneration me

166、chanisms,infringement procedures,and bias eliminations.Record keeping should be introduced as a mandatory requirement for providers of AI applications that release AI music in the UK regardless of the territory of origin of the AI application or AI creation.MMFThe MMF supports the principles of Cons

167、ent,Control and Compensation for the use of artists music in AI.We believe that alongside the requirement for licensing the music copyrights,express consent from music-makers should be sought,not assumed,by both AI companies and rightsholders for AI ingestion and generation.In addition,the UK needs

168、to introduce a strong legal backing for personality rights to ensure that individuals have control over specific usage of both their face and their voice and this cannot be assigned to corporations on a blanket basis.2829The APPG on Music inquiry into AI and music has shown how pivotal action,legisl

169、ation and education is in this area.Across the four themes of consumer protection,fair market access,voice and image likeness and international action this report has highlighted nine recommendations,underpinned by the need for a UK AI act.On consumer protection,we heard of the need for consumers to

170、 be able to make an informed decision on the music they consume and whether its generated by AI.To make this choice,AI-generated music must be labelled.For music to be labelled,tech providers must keep a record of what music their platforms are ingesting.“I actually dont think that this is the oppos

171、ite of being pro-tech.What tech companies appreciate more than anything else is certainty and predictability.They know where they stand”.Noam Shemtov on the strength of certainty On fair market access,creators choice is fundamentally important,so that the people who create and make the music we love

172、 and listen to can decide how their work is used.The APPG on Music heard about exceptions and opt-outs,and why they do not work.On voice and image likeness,the inquiry highlighted examples from the US where strong personality rights have been proposed and introduced at the federal and state level si

173、nce 2023.This showed the case that the UK should introduce a personality right to protect the individuality of creators in the UK and not fall behind our international competitors.Finally,on international action the APPG on Music considered how we can deal with these issues on a global scale,both fr

174、om the perspective of copyright and international cooperation.The suggestion of a pro-creative industries international taskforce on AI supports what we heard from international counterparts in the EU and US on the need to work closely together in the future.The APPG on Music looks forward to workin

175、g with policymakers and industry stakeholders internationally to bring these recommendations to action.“I certainly like to believe that human creativity cannot be replicated by a machine.Combining bits of everything that came before seems unlikely to result in something as truly original as impress

176、ionism or cubism or twelve-tone music”.Shira Perlmutter on the strength of human creativity Conclusion31The APPG on Music would like to thank all of the participants,individuals,and organisations who gave their time to support the inquiry.In particular,we would like to thank Florian Koempel,Burak zg

177、en,Noam Shemtov,Shira Perlmutter,Rachel Lyske and the wider DAACI team for their contributions to the virtual and in-person sessions they attended.These contributions have been invaluable to the recommendations and narrative of the wider report.The APPG on Music thanks all the MPs and Peers who atte

178、nded and engaged with the sessions.Thank you for joining,asking questions and showing an interest in this area at a pivotal time.Thank you to the APPG on Music Chair,Kevin Brennan MP,for leading on this work.The APPG on Music would also like to thank UK Music,for supporting with the sessions and hel

179、p with drafting this report.AcknowledgmentsProvided by the UK Parliament POST summary AI glossary13Algorithm A set of instructions used to perform tasks(such as calculations and data analysis)usually using a computer or another smart device.Artificial Intelligence The UK Governments 2023 policy pape

180、r on A pro-innovation approach to AI regulation defined AI,AI systems or AI technologies as“products and services that are adaptable and autonomous”.The adaptability of AI refers to AI systems,after being trained,often developing the ability to perform new ways of finding patterns and connections in

181、 data that are not directly envisioned by their human programmers.The autonomy of AI refers to some AI systems that can make decisions without the intent or ongoing control of a human.Deepfakes Pictures and video that are deliberately altered to generate misinformation and disinformation.Advances in

182、 generative AI have lowered the barrier for the production of deepfakes.Deep Learning A subset of machine learning that uses artificial neural networks to recognise patterns in data and provide a suitable output.Deep learning is suitable for complex learning tasks and has improved AI capabilities in

183、 tasks such as voice and image recognition,object detection and autonomous driving.Generative AI An AI model that generates text,images,audio,video or other media in response to user prompts.It uses machine learning techniques to create new data that has similar characteristics to the data it was tr

184、ained on.Generative AI applications include chatbots,photo and video filters,and virtual assistants.Large Language Models A type of foundation model that is trained on vast amounts of text to carry out natural language processing tasks.During training phases,large language models learn parameters fr

185、om factors such as the model size and training datasets.Parameters are then used by large language models to infer new content.While there is no universally agreed figure for how large training datasets need to be,the biggest large language models(frontier AI)have been trained on billions or even tr

186、illions of bits of data.For example,the large language model underpinning ChatGPT 3.5(released to the public in November 2022)was trained using 300 billion words obtained from internet text.See also natural language processing and foundation models.Responsible AI Often refers to the practice of desi

187、gning,developing,and deploying AI with certain values,such as being trustworthy,ethical,transparent,explainable,fair,robust and upholding privacy rights.Transformers Transformers have greatly improved natural language processing,computer vision and robotic capabilities and the ability of AI models t

188、o generate text.A transformer can read vast amounts of text,spot patterns in how words and phrases relate to each other,and then make predictions about what word should come next.This ability to spot patterns in how words and phrases relate to each other is a key innovation,which has allowed AI mode

189、ls using transformer architectures to achieve a greater level of comprehension than previously possible.See also artificial neural networks and generative adversarial networks.Glossary of Terms References can be found on page 33331https:/hansard.parliament.uk/commons/2023-02-01/debates/7CD1D4F9-7805

190、-4CF0-9698-E28ECEFB7177/Artifi-cialIntelligenceIntellectualPropertyRights2https:/ 3Full tables can be found on Whitestone Insights website.4https:/www.ifpi.org/music-fans-worldwide-believe-human-creativity-essential-in-time-of-ai/5Clement-Jones,T.and Darling,C.(2024).Living with the Algorithm:Servan

191、t or Master?:AI Governance and Policy for the Future.Unicorn Publishing Group.6The Creative Artists Agency submitted written evidence to the inquiry.7https:/www.supremecourt.uk/cases/uksc-2021-0201.html8https:/www.bailii.org/ew/cases/EWCA/Civ/2023/1354.html 9https:/eur-lex.europa.eu/legal-content/EN

192、/TXT/?uri=CELEX:62017CJ0683 10 https:/ https:/www.copyright.gov/comp3/docs/compendium.pdf n.b.appeal12 Jonathan Coote,Music and IP Lawyer at Bray&Krais,submitted written evidence to the inquiry.13 https:/post.parliament.uk/artificial-intelligence-ai-glossary/ReferencesAs part of the inquiry,the APPG

193、 welcomed written evidence from the public.The following questions were shared with the public to answer:1.Are you currently using any AI tools or technology to help you create music?If yes,please explain how you use AI in your music-making process.If no,why not?2.To what extent do you agree or disa

194、gree with the idea that legal ownership should only apply to creative works made by humans?3.Do you think artists should have the choice to decide if their music is used to train AI systems?4.How important is it,in your opinion,for tech companies to keep a record of the music they use to train AI sy

195、stems?5.Should music created with the help of AI be clearly labelled as such?If so,how should this labelling be implemented?6.Do you believe a new legal right should be established to safeguard the personality,image,and likeliness of songwriters,artists,and performers,to protect from things such as

196、visual and audio deepfakes?The APPG on Music thanks the individuals and organisations that took the time to respond.In summer 2023,UK Music published five key principles that the government should adopt in relation to AI and music.This is available here:https:/www.ukmusic.org/wp-content/uploads/2023/07/UK-Music-Policy-Position-Paper-on-Artificial-Intelligence.pdf.Annex